Skip to content

Add Windows ARM64 GHA runner #31867

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 1 commit into
base: main
Choose a base branch
from

Conversation

vask2108
Copy link

@vask2108 vask2108 commented Aug 1, 2025

Reference Issues/PRs

Fixes issue : #30567

What does this implement/fix? Explain your changes.

This PR will add CI work to create Windows on ARM64(WoA) wheel.

Any other comments?

Please note SciPy Windows on Arm64 (WoA) is recently committed and SciPy wheel will take sometime to appear in PyPI. Currently I have to create a temporary patch to pick it from pypi.anaconda.org. I will remove this temporary solution soon, once PyPI hosts SciPy wheel. Please see the comment here ( scipy/scipy#23122 (comment) )

Please also note WoA GHA runners are new (available from April 2025) and docker is not available yet. So docker is skipped for WoA now.

Copy link

github-actions bot commented Aug 1, 2025

✔️ Linting Passed

All linting checks passed. Your pull request is in excellent shape! ☀️

Generated for commit: 721947f. Link to the linter CI: here

@vask2108 vask2108 changed the title Add Winows ARM64 GHA runner Add Windows ARM64 GHA runner Aug 1, 2025
Copy link
Member

@adrinjalali adrinjalali left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

need a commit with [cd build] message to test the wheel builds

@adrinjalali
Copy link
Member

@vask2108 please never force push here. It makes reviews harder, and we squash and merge anyway. This can be put as a draft PR since I don't think we should really look at it much until our dependencies have all the wheels available first.

@vask2108
Copy link
Author

vask2108 commented Aug 4, 2025

@adrinjalali
Thank you very the review.

As it was cosmetic change, I did a the force push. But I can understand the inconvenience, I will definitely avoid next time.

What I understand from your response is that, you would like to wait for SciPy (dependency) wheel to be available in PyPI to proceed further with this PR? Until then, you consider this as draft PR.

I don't think any other dependency wheel is pending apart from SciPy

Is my understanding is correct?

Thank you.

@adrinjalali
Copy link
Member

Yes, that sounds about right.

Copy link
Member

@ogrisel ogrisel left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for the PR. I took a look at the diff and scanned the CI outputs, and it looks good to me.

But I would also rather wait for SciPy to publish a stable release with Windows ARM64 wheels to avoid having to rely on the sed-based hackish workaround in our CI config.

However, if a downstream project of scikit-learn would like to have access to nightly Windows ARM64 builds before that happens, I am fine with merging that PR earlier.

@@ -173,6 +189,24 @@ jobs:
with:
miniforge-version: latest

# PyPI not yet hosting SciPy wheel as WoA PR of SciPy recently merged.
# This workaround will be removed once PyPI hosts SciPy wheel.
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggested change
# This workaround will be removed once PyPI hosts SciPy wheel.
# TODO: remove this workaround once PyPI hosts SciPy wheel.

@lesteve
Copy link
Member

lesteve commented Aug 5, 2025

+1 for waiting for a scipy release with Windows arm wheels. According to scipy/scipy#23122 (comment) a scipy release with Windows arm wheels may happen some time in September 2025.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants